DeepSeek crashes US tech stocks

Summary

DeepSeek, a low-cost Chinese AI chatbot built on open-source models and reportedly costing around $6 million to develop, has become the most popular free app in the US. This development triggered a sharp sell-off in tech stocks, with Nvidia losing over $500 billion in market value in a single day.

Why it matters

DeepSeek directly challenges the economic rationale behind the $500 billion in US AI infrastructure investments and the chip-centric valuations of companies like Nvidia and ASML. If frontier-level performance can be achieved at a fraction of the assumed cost, the entire investment thesis for AI hardware and cloud computing is open to repricing.

Highlights

• Nvidia dropped by more than 16% (−$500 billion in market value); ASML fell by over 10%; Siemens Energy lost 21% of its value in a single session.
• The claimed $6 million training cost is being questioned by other AI experts. However, this figure forms the basis of the market's fear.
• Citi analysts noted that the US's access to advanced chips remains a structural advantage. This limits the narrative that 'China has won.'

Critical take

The article relies too heavily on the unverified $6 million cost figure and a dramatic 'Sputnik moment' framing. The fact that other experts question this number is only briefly mentioned. This presents readers with a misleading cost comparison. Additionally, the article fails to sufficiently emphasize that DeepSeek's claim of matching the performance of OpenAI's latest models is based on its own assertions and that its long-term sustainability under the tightened US chip export restrictions has not yet been tested.